Anaconda a Real-time Control-flow/data-flow Hybrid Processor

Typically real-time applications are written in a multi-threaded language and are executed on single threaded control-ow (von Neumann) processors. A software scheduler is used to switch between the application's multiple threads, thereby simulating a multi-threaded processor. Unfortunately software scheduling is problematic if real-time constraints are to be guaranteed and therefore static (predetermined) scheduling is often used even though this prohibits sporadic tasks. Control-ow processors, being inherently single threaded, are unable to use concurrency to tolerate latency. Therefore, the faster control-ow processors become, the more they rely on caches to reduce memory access latency. Unfortunately , caches are also a signiicant source of temporal complexity, especially when scheduling is involved. Data-ow processors are inherently multi-threaded and are able to tolerate la-tency. Unfortunately, current data-ow processors have very simplistic sched-ulers which prevent real-time requirements from begin met. Furthermore, assignment , and thus a sense of history which is vital for many real-time applications , is problematic. Control-ow/data-ow hybrids have been proposed in the past but they do attempt the challenge of scheduling for real-time. This report presents a possible real-time control-ow/data-ow hybrid processor | Anaconda.